Taking feature requests!

[Joshuaalbert]

What would you like to see in JAXNS? What would make your problem solving easier?

1. Do you want better model representation (the probabilistic programming framework side)?
2. Do you want to see more examples combining ML and Bayesian computation?
3. Do you want better use of gradients (currently experimental)?
4. Do you want better default parameters?
5. Do you want better distributed solving (compute cluster problem solving)?
6. Do you want better solving super hard problems full of degeneracies?
7. Do you want to use normalising flows to incrementally repartition the problem so that it uses fewer likelihood evals, at the cost of overhead compute?
8. Do you want to see more special priors? Which ones?
9. Do you want more standardised benchmarks to assess accuracy in more problem settings?
10. Do you want better evidence-EM (currently experimental)?
11. Do you want better global likelihood maximisation (currently experimental)?
12. Something else?